I-STROKE SCORE: A RISK PREDICTION SCORING TOOL FOR CEREBROVASCULAR DISEASE AMONG TYPE 2 DIABETES MELLITUS PATIENTS

About the Project

Cerebrovascular disease (CeVD) is a leading cause of death and disability among people with Type 2 Diabetes Mellitus (T2DM) in Malaysia, yet most stroke risk tools were developed in Western populations and depend on investigations or computing infrastructure that primary care clinics lack. The I-Stroke Score was created as a transparent, point-based and AI-calibrated risk prediction tool to identify T2DM patients at higher CeVD risk during routine primary care. Data from 20,338 patients in the Pulau Pinang National Diabetes Registry (2019 to 2024) were analysed across three temporal phases: development, temporal internal validation and temporal external validation. Logistic regression with backward elimination produced an eight-predictor score using routinely recorded variables: age, sex, diabetes duration, body mass index, hypertension, chronic kidney disease stage, ischaemic heart disease and district. Patients are classified into Low, Medium and High risk tiers. Seven machine learning (ML) algorithms, including a neural network, and an ensemble model were benchmarked using the same predictors. The score achieved areas under the curve of 0.681, 0.663 and 0.630 across the three phases, with observed CeVD prevalence rising from the Low to the High risk tier, and no ML model significantly outperformed it in any phase. When calibration drifted in the most recent cohort, logistic recalibration (Platt scaling) removed statistical evidence of miscalibration (Hosmer-Lemeshow p = 0.261) without changing the points table. The I-Stroke Score is live as a free web calculator, and the calculator and scoring method are copyright registered with MyIPO (CRLY2025W02431). It has clear commercialisation potential: free for public clinics, it can be integrated into clinic electronic record systems and offered to private general practice networks as a subscription-based risk screening module, while recalibration at each new registry audit creates a sustainable service model adaptable to other states and the wider ASEAN region. In line with the theme "Inventing the Future, Empowering the Nation", the I-Stroke Score turns data that clinics already collect into low-cost, scalable stroke risk stratification for resource-limited primary care.